Source code for: Virtual Immunohistochemical Staining for Pathological Complete Response Prediction in Triple-Negative Breast Cancer
This record contains the source code supporting both the article (submitted to Bioinformatics, 2026) and the master's dissertation (PPGCC/UFMG, 2026) entitled "Virtual Immunohistochemical Staining for Pathological Complete Response Prediction in Triple-Negative Breast Cancer". The project predicts pathological complete response (pCR) in triple-negative breast cancer (TNBC) by combining H&E images with Ki-67 and PHH3 stains (real or virtually generated) through a biomarker-based spatial attention mechanism. Contents: bcs-kedro/ - Kedro project implementing the full pCR prediction pipeline: image preprocessing, generation of biomarker-based spa
Resource information
- Format
- zip
- Original repository
- 23788133
- DOI
- 10.5281/zenodo.23788133
- Credit basis
- Creators in Zenodo record metadata
- Access
- Open files available from the original Zenodo deposit
- Last checked
- Nov 5, 2026
- License
- mit-license
- Last record update
- Nov 5, 2026
About this resource
This record contains the source code supporting both the article (submitted to Bioinformatics, 2026) and the master's dissertation (PPGCC/UFMG, 2026) entitled "Virtual Immunohistochemical Staining for Pathological Complete Response Prediction in Triple-Negative Breast Cancer". The project predicts pathological complete response (pCR) in triple-negative breast cancer (TNBC) by combining H&E images with Ki-67 and PHH3 stains (real or virtually generated) through a biomarker-based spatial attention mechanism. Contents: bcs-kedro/ - Kedro project implementing the full pCR prediction pipeline: image preprocessing, generation of biomarker-based spatial attention maps, and a ResNet-based classifier with attention. virtual_staining/ - models for generating Ki-67 and PHH3 stains from H&E images, including CycleGAN, conditional GAN (cGAN), and diffusion-based approaches. Requirements: Python 3.8+, dependencies listed in requirements.txt. Training is GPU-optimized.ob scripts are configured for SLURM clusters. Installation and execution instructions are provided in README.md. Trained models: Hugging Face | Digital Pathology Models Source repository: Github | PCR Classification Repository Main author : Henrique Colonese Main contributor : Daniel Ayala